Transcript
Page 1: We Can Hear You with Wi-Fi !

We Can Hear You with Wi-Fi !

Guanhua Wang

Yongpan Zou, Zimu Zhou, Kaishun Wu, Lionel M. Ni

Hong Kong University of Science and Technology

Page 2: We Can Hear You with Wi-Fi !

Advanced Research in ISM band

• Localization

Page 3: We Can Hear You with Wi-Fi !

Advanced Research in ISM band

•Gesture Recognition

Page 4: We Can Hear You with Wi-Fi !

Advanced Research in ISM band

•Object Detection

Full Duplex Backscatter. In hotnets 2013

Page 5: We Can Hear You with Wi-Fi !

• Localization

•Gesture recognition

•Object Classification

They enable Wi-Fi to “SEE” target objects.

Advanced Research in ISM band

Page 6: We Can Hear You with Wi-Fi !

Can we enable Wi-Fi signals to HEAR talks?

Page 7: We Can Hear You with Wi-Fi !

Can we enable Wi-Fi signals to HEAR talks?

Page 8: We Can Hear You with Wi-Fi !

What is WiHear?

Page 9: We Can Hear You with Wi-Fi !

“Hearing” human talks with Wi-Fi signals

Hello

Non-invasive and device-free

Page 10: We Can Hear You with Wi-Fi !

Hearing through walls and doors

I am upset.

Understanding complicated human behavior (e.g.

mood)

Page 11: We Can Hear You with Wi-Fi !

Hearing multiple people simultaneously

MIMO Technology

Easy to be implemented in commercial Wi-Fi products

Page 12: We Can Hear You with Wi-Fi !

How does WiHear work?

Page 13: We Can Hear You with Wi-Fi !

WiHear Framework

Feature Extraction

Classification & Error Correction

Learning-based Lip Reading

LaptopLaptop

PeoplePeople

APAP

MIMO Beamforming

Vows and consonants Filtering

Wavelet Transform

Remove Noise

Mouth Motion Profiling

Partial Multipath Removal

Segmentation

ProfileBuilding

Page 14: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 15: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 16: We Can Hear You with Wi-Fi !

Locating on Mouth

APAP

PeoplePeople

LaptopLaptop

T1

T2T3

T3

Page 17: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 18: We Can Hear You with Wi-Fi !

Filtering Out-Band Interference

•Signal changes caused by mouth motion: 2-5 Hz

•Adopt a 3-order Butterworth IIR band-pass filterCancel the DC componentCancel wink issue (<1 Hz)

Cancel high frequency interference

The impact of wink (as denoted in the dashed red box).

Page 19: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 20: We Can Hear You with Wi-Fi !

Partial Multipath Removal

•Mouth movement: Non-rigid

• Covert CSI (Channel State Information) from frequency domain to time domain via IFFT

•Multipath removal threshold: >500 ns

• Convert processed CSI (with multipath < 500ns) back to frequency domain via FFT

The multipath threshold value can be adjusted to achieve better performance

Page 21: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 22: We Can Hear You with Wi-Fi !

Mouth Motion Profiling

• Locating on Mouth

• Filtering Out-Band Interference

• Partial Multipath Removal

•Mouth Motion Profile Construction

• Discrete Wavelet Packet Decomposition

Page 23: We Can Hear You with Wi-Fi !

Discrete Wavelet Packet Decomposition• A Symlet wavelet filter of order 4 is selected

Page 24: We Can Hear You with Wi-Fi !

WiHear Framework

Feature Extraction

Classification & Error Correction

Learning-based Lip Reading

LaptopLaptop

PeoplePeople

APAP

MIMO Beamforming

Vows and consonants Filtering

Wavelet Transform

Remove Noise

Mouth Motion Profiling

Partial Multipath Removal

Segmentation

ProfileBuilding

Page 25: We Can Hear You with Wi-Fi !

Lip Reading

• Segmentation

• Feature Extraction

• Classification

• Context-based Error Correction

Page 26: We Can Hear You with Wi-Fi !

Lip Reading

• Segmentation

• Feature Extraction

• Classification

• Context-based Error Correction

Page 27: We Can Hear You with Wi-Fi !

Segmentation

• Inter word segmentationSilent interval between words

• Inner word segmentation Words are divided into phonetic events

Page 28: We Can Hear You with Wi-Fi !

Lip Reading

• Segmentation

• Feature Extraction

• Classification

• Context-based Error Correction

Page 29: We Can Hear You with Wi-Fi !

Feature Extraction• Multi-Cluster/Class Feature Selection (MCFS) scheme

Page 30: We Can Hear You with Wi-Fi !

Lip Reading

• Segmentation

• Feature Extraction

• Classification

• Context-based Error Correction

Page 31: We Can Hear You with Wi-Fi !

Lip Reading

• Segmentation

• Feature Extraction

• Classification

• Context-based Error Correction

Page 32: We Can Hear You with Wi-Fi !

Extending To Multiple Targets

•MIMO: Spatial diversity via multiple Rx antennas

• ZigZag decoding: a single Rx antenna

Page 33: We Can Hear You with Wi-Fi !

Implementation

Floor plan of the testing environment. Experimental scenarios layouts. (a) line of sight; (b) non-line-of-sight; (c) through wall Tx side; (d) through wall Rx side; (e) multiple Rx; (f) multiple link pairs.

Page 34: We Can Hear You with Wi-Fi !

Vocabulary

• Syllables: [æ], [e], [i], [u], [s], [l], [m], [h], [v], [ɔ], [w], [b], [j], [ ʃ ].

•Words: see, good, how, are, you, fine, look, open, is, the, door,

thank, boy, any, show, dog, bird, cat, zoo, yes, meet, some, watch, horse, sing, play, dance, lady, ride, today, like, he, she.

Page 35: We Can Hear You with Wi-Fi !

Automatic Segmentation Accuracy

Automatic segmentation accuracy for (a) Inner-word segmentation

on commercial devices (b) Inter-word segmentation on commercial devices (c) Inner-word segmentation on USRP(d) Inter-word segmentation on USRP

Page 36: We Can Hear You with Wi-Fi !

Classification Accuracy

Page 37: We Can Hear You with Wi-Fi !

Training Overhead

Page 38: We Can Hear You with Wi-Fi !

Impact of Context-based Error Correction

Page 39: We Can Hear You with Wi-Fi !

Performance with Multiple Receivers

Example of different views for pronouncing words

Page 40: We Can Hear You with Wi-Fi !

Performance for Multiple Targets

Performance of multiple users with multiple link pairs.

Performance of zigzag decoding for multiple users.

Page 41: We Can Hear You with Wi-Fi !

Through Wall Performance

Performance of two through wall scenarios. Performance of through wall with multiple Rx.

Page 42: We Can Hear You with Wi-Fi !

Resistance to Environmental Dynamics

Waveform of a 4-word sentence without interference of ISM band signals or irrelevant human motions

Impact of irrelevant human movements interference

Impact of ISM band interference

Page 43: We Can Hear You with Wi-Fi !

Conclusion

• WiHear is the 1st prototype in the world, trying to use Wi-Fi signal to sense and recognize human talks.

• WiHear takes the 1st step to bridge communication between human speaking and wireless signals.

• WiHear introduces a new way so that machine can sense more complicated human behaviors (e.g. mood).

Page 44: We Can Hear You with Wi-Fi !

Thank you for your listening !

Page 45: We Can Hear You with Wi-Fi !

Questions ?

Guanhua Wang

[email protected]

WiHearWe Can Hear You With Wi-Fi !


Recommended